demeo taxonomy: categorization of asteroids in the near
TRANSCRIPT
DeMeo Taxonomy: Categorization of Asteroids in the Near-Infrared
By
Francesca E. DeMeo
B.S., Earth, Atmospheric and Planetary Sciences (2006)B.S., Physics (2006)
Massachusetts Institute of Technology
Submitted to the Department of Earth, Atmospheric and Planetary Sciencesin Partial Fulfillment of the Requirements for the Degree of
Master of Science in Earth, Atmospheric, and Planetary Sciences
at the
Massachusetts Institute of Technology
June 2007
C 2007 Massachusetts Institute of Technology.All rights reserved.
AuthorDepartment of Earth, Atmospheric and Planetary Sciences
June 2007
Certified byProfessor kichPf . Binzel
Thesis Supervisor
Accepted by
MASSACHUSETTS INSTITUTEOF TECHNOLOGY
AUG 3 0 2007
LIBRARIES
J(4aria T. ZuberE.A. Griswold Professor of Geophysics
Head, Department of Earth, Atmospheric & Planetary Sciences
ARCHN*ES
Table of Contents
A bstract................................ ................................................ 3......Acknowledgements ............................... ........................................ 41. Introduction ........................................................................... .. 52. Observations and Data Reduction ....... ......... ............................ .. 9......3. Background on Classification ......................................... ......... 11
3.1 Thoughts on Classification............................................. 113.2 Previous Classification Systems........................................ 123.3 Process to New Classification ............................................ 13
3.3.1 PCA Preparation ............................................... 133.3.2 What Principal Components Tell ............................... 163.3.3 Choosing the Right Number of Principal Components....... 173.3.4 Notation: The difference between PC1 and PC ' ........... 20
4. The Taxonom y ............................................................................. . 215. Conclusion ................................................................................ 39R eferences.................................... .............................................. 40Appendix A: Bus-DeMeo Class Evolution .............................................. 43Appendix B: Observations...............................................44Appendix C: Principal Component Eigenvectors......................................48Appendix D: Flow Chart ............................ ..................................... 49Appendix E: New Labels to Data ......... .......... ........................................ 55
DeMeo Taxonomy: Categorization of Asteroids in the Near-InfraredBy
Francesca E. DeMeo
Submitted to theDepartment of Earth, Atmospheric and Planetary Sciences
June 2007
In Partial Fulfillment of the Requirements for the Degree ofMaster of Science in Earth, Atmospheric, and Planetary Sciences
Abstract
This work presents the DeMeo taxonomy, an asteroid taxonomy with 24 classes based onPrincipal Component Analysis of spectral data over the visible and near-infraredwavelengths, specifically the 0.45 to 2.45 micron range. Principal Component Analysiswas used by both Tholen (1984) and Bus (1999) to create taxonomies on visible data.There is no pre-existing taxonomic system for the entire suite of asteroid characteristicsbecause only in the current decade has spectral data collection become available in thenear-infrared for asteroids down to relatively faint (V= 17) limiting magnitudes. (Rayneret al. 2003) With a larger data range, which includes important absorption features at oneand two microns suggesting the presence of minerals, there is a need for an extendedsystem to encompass this range of information. In this work we explain the process ofcreating the taxonomy, the method for finding an object's taxonomic class under thissystem, and present spectral types for the 365 objects that were used to create the system.
Thesis Supervisor: Richard P. BinzelTitle: Professor of Planetary Science
Acknowledgements
I am grateful for the help and advice from many colleagues and students. I thank
Bobby Bus for all his guidance and the expertise he provided having created a previous
taxonomic system. Thanks to Pierre Vernazza for all the mineralogical background and
input he contributed which helped me understand the distinctions between spectra.
Thanks to Cristina Thomas for always being available to help me with any problems or
questions and for always taking the tough night shift during observing runs. I thank
Shaye Storm for being a great UROP student and for helping me reduce and organize the
data for this work. I would also like to thank many others who have taught me much and
helped me along the way. Thanks Andy Rivkin, Mike Person, Steve Slivan, Jim Elliot,
Jack Wisdom, Elisabeth Adams, Mark Avnet, and Susan Kern. I thank my family for
everything.
I also owe huge thanks to my advisor Professor Rick Binzel who has guided me
through not just my Masters thesis, but also my senior thesis. Thank you Rick for all the
opportunities you have provided for me, the chance to visit the IRTF on Mauna Kea and
Magellan in Chile, to travel to and present at conferences, to research in France and meet
colleagues. Thank you for always taking the time to help and answer questions, for going
out of your way to make sure you're available to help, for always being patient, and for
always being accommodating for the projects I undertake and plans I make outside of my
research. And by the way, that is a sweet Earth.
1. Introduction
Classification is important to all branches of sciences. As soon as a large amount
of data are available, it is useful to group it into categories with similar characteristics for
organization, comparison, and better comprehension.
The first asteroid taxonomy was created by Chapman et al. (1975). The
taxonomy has grown and evolved over time to account for more and better data. Asteroid
scientists started by characterizing asteroids with a few colors, and then many colors as
technology improved. Some taxonomies incorporated albedo, and eventually they used
spectra. Taxonomies were created by Tholen (1984) and most recently by Bus (1999).
All current asteroid taxonomies are based on the visible wavelength range approximately
0.4 to 0.9 microns because data were most available over those wavelengths. Recently,
with the creation of the SpeX instrument on the IRTF a growing library has been created
of near-infrared spectral data. Gaffey et al. (1993) created an S-complex taxonomy based
on near-infrared data, but there is no pre-existing taxonomic system for the entire suite of
asteroid characteristics. From Figures 1 and 2 we can see examples of the need for a new
taxonomy because of the divergence in the near-infrared of a single well-constrained
visible class. A classification system that extends to the near-infrared is also important
because significant mineral absorption features are present at one and two microns
suggesting presence of pyroxene and olivine. Even though the taxonomy presented in
this work is not connected to mineralogy it is hoped that future work would allow
mineralogic interpretation of taxonomic classes.
A Types.... . . . .. . . . . . . . . . . .. . . . . . . . . ............. I ....... ..... .... ........ ............... ........... ...... ........................... ....................... . ..1 -1 1
.13i 1.5
1I·
0.5
.2468 289A 863x446x 2715* 4142
S5841~~
1.5
Wavelength (microns)
Figure 1: Bus Taxonomy A-Types. All spectra behave similarly in the visible and aredesignated A-Types in the Bus system, but have very different behaviors in the near-infrared.
_ __ __. __ · _I _ ___1~1 _~__
L-Types0.5
2
0.5-
a.0 0.5 1 1.5 2 2.5 3
Wavelength (microns)
Figure 2: Bus Taxonomy L-Types. This plot shows L-types under the Bus classificationsystem. Notice how well they match in the visible range, but diverge in the near-infrared.Some have stronger 1 micron absorptions, some have high slopes or even negative slopesafter one micron.
This work presents the DeMeo taxonomy, an asteroid taxonomy based on
Principal Component Analysis of spectral data over the visible and near-infrared
wavelengths, specifically the 0.45 to 2.45 micron range. The aim of this new taxonomy
is to have classification parameters for the entire wavelength range, but not to create an
entirely new system. Principal Component Analysis (PCA) is a technique for reducing
the dimensionality of a dataset to contain most of the information in a few principal
components. It is a set of linear transformations changing the coordinate system
according to the greatest variance. The greatest variance is along the dimension
S42
A 908x729x 387
* 460
. 2448
OC1·111111_1~
described by Principal Component 1 and the second greatest variance is Principal
Component 2. PCA is used widely in other applications such as data compression, but
was not applied to the field of asteroid taxonomy until Tholen (1984).
When creating this taxonomy there were seven possible outcomes for changes to
each taxonomic class under the Bus system. Two classes could be indistinguishable over
the entire wavelength range and merge into one class. One class could be degenerate and
branch off into two (or more) separate classes. A class may need to be created if spectra
exist that do not fit within the past framework. We found that one new class needed to be
created, the Sv class, comprised of two objects that represented an intermediate class
between S and V. The name of a class may need to be changed if the current notation
seems misleading. Some classes may need to be redefined. For example the names Sr
and Sa were conserved but the objects in those classes were largely redefined. There is
the possibility that one class may correlate directly as near-infrared data are included and
no change in that class occurs. Finally, it is possible that a class may remain the same,
but a notation is added. In this taxonomy a "w" notation was used to indicate high slope
and is explained in Section 4. Figure 3 is a diagram of the possible outcomes of
extending the classes. See Appendix A for the progression from the Bus to DeMeo
taxonomy.
Figure 3: Possible outcomes of extending taxonomic classification to the near-infrared. 1.Two classes from the Bus system could merge to one class. 2. One class could divergeinto two distinct classes. 3. A class that did not previously exist may be created. 4. Apreviously existing class may be relabeled with a different name. 5. A class may keep itslabel but its definition may change. 6. One class could correlate directly from the Bus toDeMeo taxonomies. 7. An additional notation to may be added within a class.
2. Observations and Data Reduction
Near-infrared spectral measurements from 0.8 to 2.5 microns were obtained using
the SpeX instrument on the 3 meter NASA Infrared Telescope Facility (IRTF) located on
Mauna Kea, Hawaii. As described in DeMeo and Binzel (2007), when possible, objects
and standard stars were observed near the meridian to minimize their differences in
airmass. Frames were taken so that the object was alternated between two different
positions (usually noted as the 'A' and 'B' positions) on a 0.8 x 15 arcsecond slit aligned
north-south. Solar-type standard stars were observed to divide the solar spectrum from
1. Merge: S S
2. Diverge: Ld L
D
3. Create: - Sv
4. Relabel: Sk Sq
5. Redefine: Sr - Sr
6. No Change: O 0O
7. Add a Notation: S -0 Sw
the asteroid spectra, with the result being reflectance spectra. Our primary solar analog
standard stars were 16 Cyg B and Hyades 64. Additional solar analog stars were utilized
around the sky, having comparable spectral characteristics. Two to three sets of eight
images per set were taken for each object, with individual image exposures typically
being 120 seconds. The total integration time for each of these objects therefore ranged
from 30 to 120 minutes.
Reduction was performed using a combination of routines within the Image
Reduction and Analysis Facility (IRAF), provided by the National Optical Astronomy
Observatories (NOAO) (Tody 1993), and Interactive Data Language (IDL). We use a
software tool called "autospex" to streamline reduction procedures outlined by S. J. Bus
(personal communication). Autospex writes macros containing a set of IRAF (or IDL)
command files that are then executed by the image processing package. Autospex
procedures operate on a single night at a time, with the opportunity for the user to inspect
and verify the results at each stage. Briefly, autospex writes macros that: trim the images
down to their useful area, create a 'bad pixel map' from flat field images, flat field correct
all images, perform the sky subtraction between AB image pairs, register the spectra in
both the wavelength and spatial dimensions, co-add the spectra for individual objects,
extract the 2-D spectra from co-added images, and then apply the final wavelength
calibration. Using IDL, an absorption coefficient is determined for each object and star
pair that best minimizes atmospheric water absorption effects for that pair. This
coefficient correction is most important near 1.4 and 2.0 microns, locations of major
absorption bands due to telluric H20. Because we operate in survey mode with our
observing time heavily weighted toward new objects rather than on standard stars, our
telluric corrections are not perfect. Users of these data therefore must exercise caution in
basing significant scientific conclusions on any unusual spectral features at these telluric
wavelengths, which may be the result of imperfect correction. The last IDL step averages
all the object and standard pairs to create the final reduced spectrum for each object.
All visible wavelength spectra (usually 0.4 to 0.9 microns) were taken from the
Small Main Belt Asteroid Spectroscopic Survey (SMASS II) data set (Bus & Binzel
2002). Our sample was comprised of 365 objects with both visible and near-IR data. For
a table of observations see Appendix B.
3. Background on Classification
3.1 Thoughts on Classification
Classification systems tend to emerge as soon as there are enough data to form
groups. It is a tool used to organize information making it easier to compare, find holes or
missing information, and makes it easier to communicate about data features.
Data on asteroids come from three sources: spacecraft missions, meteorites, and
telescope observations. Spacecraft missions provide the most detailed information about
asteroids, however, they only reach a limited number of targets. Meteorites provide a
great opportunity for detailed study in a laboratory, but their links to specific asteroids
and even general asteroid classes are nonexistent or weak. The only known connection
between a meteorite and its parent asteroid are the HED meteorites to Vesta
(Consolmagno & Drake 1977, Binzel & Xu 1993) Telescope observations provide data
including brightness, albedo, and spectral reflectance on a large number of asteroids
making it the best basis for classification of many objects.
The purpose of this classification system is to characterize the features present in
the spectrum. These features may represent mineralogy, grain size, weathering, or other
properties of or effects on an asteroid, however the taxonomy is not meant to suggest or
confirm anything more than the features visible in the spectrum.
3.2 Previous Classification Systems
Asteroid taxonomy has evolved significantly over the past few decades, as more
and higher quality data have become available. Starting in the 1950's UBV photometry
observations enabled the separation of asteroids into two groups noted by Wood and
Kuiper (1963) and Chapman et al. (1971). Using spectrophotometry and albedo
measurements Chapman et al. (1975) created the first letter-based asteroid taxonomy.
"C" represented carbonaceous, "S" was for stony, and "U" for unusual objects. Tholen
Taxonomy was the most widely used when created with eight classes of spectral types,
plus classes for three unusual objects. The Barucci Taxonomy, meant to expand on the
Tholen taxonomy was based on G-mode analysis. (Tholen & Barucci 1989, Barucci et
al. 1987). The most recent taxonomy was created by Bus using 1189 objects. It is
feature-based taxonomy also using PCA (Bus 1999, Bus & Binzel 2002). The Bus system
further separated the C, S, and X complexes.
Only in the current decade has spectral data collection become available in the
near-infrared for asteroids down to relatively faint (V= 17) limiting magnitudes. (Rayner
et al. 2003) With a larger data range, which includes important absorption features, there
is a need for an extended system to encompass this range of information. We chose not to
invent an entirely new system, but instead strove to stay as consistent as possible with
past visible taxonomies and notation.
3.3 Process To New Classification
3.3.1 PCA Preparation
Bus chose three prototype asteroids for each of his taxonomic classes that best
exemplify the characteristics of the class. The process of creating a new taxonomy began
with the evaluation of the Bus prototypes over the entire wavelength range. Some of
these prototypes had to be rearranged and redesignated. Asteroid 150, for example, was a
Cb under the Bus system, but looks much more like a C over the entire range. Asteroid
24 was aB and is now a C. 42 was an L and is now a K. 606 was a K and was
redesignated to L. 1904 and 5111 were R-types and are now V. Even though there were
many changes, most were within subclasses or closely related classes. This exercise was
mainly to observe how each class behaved in the near-infrared and which classes were
more or less well-defined. Because the new system has many more features present in the
near-infrared than the visible it is natural that some objects that looked similar in the
visible diverge in the near-infrared.
Next, to become more familiar with the data, we assigned a taxonomic type or
complex by visually assessing spectral features and plotting it against prototypes. This
was useful because when plotting objects in PCA space we could compare how objects
grouped when labeled under the Bus system with what we believed they looked like in
the visible and near-infrared range.
We then plotted all near-infrared data in a "splinefit" program which creates a list
of data points that follow the features of the spectrum. This step smoothes the spectra,
creating a good "best fit" curve to the data. This reduces the risk of noise or missing data
points influencing the PCA. We sampled the region 0.45 to 2.45 microns and recorded
points at increments of 0.05 resulting in 41 channels.
The "splined" data were then normalized at 0.55 microns and the slope was
removed from the data and recorded. With our normalized data we simply took a linear
regression line of each set of data. Because this is the most prominent feature of the
spectra we can remove it from the data before performing principal component analysis
thus making PCA more sensitive to other features, and therefore more effective. We
divide all the data points by the average slope. The remaining data are a horizontal
spectrum with residuals (including absorption features) above and below the horizontal.
We now have data with slope removed having 41 channels normalized at 0.55 microns.
Because each spectrum has a value of 1 at 0.55, that channel provides no new
information and so was removed from the data set to make PCA more effective. We
therefore input 40 channels per object into PCA
To verify that we could remove the slope before running PCA, we ran PCA on the
data without the slope removed. The first principal component (PC1) of PCA before
removing slope was compared to the slope we remove directly from the splined data. In
Figure 4, a plot of PC1 versus slope, it is clear that the two are linearly correlated, thus
PC1 corresponds to slope and we are justified in removing it first.
1.0
U)
0.5
0.0
-4 -Z 0 2 4 6 B
PC1
Figure 4: Plot of PCi versus Slope. PCI is the first principal component from the dataset that did not have the slope removed. It is clear from the plot that PCI and slope arelinearly related and it is thus safe to remove the slope before running PCA to make PCAmore sensitive to other features.
Using MATLAB, we performed Principal Component Analysis (PCA) on the
splined files with slopes removed. PCA is a method of reducing the dimensionality of a
data set. PCA involves coordinate transformations to minimize the variance. The first
transformation rotates the data to maximize variance along the first axis, known as
Principal Component 1' (PC1'), the second axis is the second Principal Component
(PC2'). The first few principal components contain the majority of the information. For a
more thorough explanation of PCA and why it is useful for asteroid taxonomy refer to
Tholen (1984) and Bus (1999).
UI
U U
r~ *r
a
U
U
UI1 . I . .tI .~
3.3.2 What Principal Components Tell
By plotting the principal components against one another we can see how
different groups separate. PCA was used to create the Tholen and Bus taxonomies. We
chose to use covariance instead of correlation for the PCA. Bus (1999) suggested the use
of the covariance matrix of unscaled variances because they may be useful to distinguish
spectra. Principal Component Analysis tells information about specific parts of the
spectrum. It is difficult to determine what exact feature each principal component refers
to because so many features exist over the visible to near-infrared wavelength. However,
by looking at the eigenvectors we can determine over which wavelength range each
principal component is most sensitive. Figure 5 is a plot of sensitivity of the eigenvectors
throughout the range. This figure shows that PC4' and PC5' are more sensitive to
features at the ends of the spectrum, while the others are more sensitive to features in the
middle. See Appendix C for the eigenvectors. Wavelengths with higher magnitudes
dominate that principal component using positive or negative weights to separate in
opposite directions within that principal component space.
Principal Component Eigenvectors
4--0 PC1'
-- PC2'-- PC3'
P-- PC4'
- PC5'
0.0 0.5 1.0 1.5 2.0 2.5 3.0
Wavelength
Figure 5: Sensitivity of Principal Components over wavelength range. High positive ornegative values indicate that a principal component is more sensitive to information overthat region. This figure shows that PC4' and PC5' are more sensitive to features at theends of the spectrum, while the others are more sensitive to features in the middle.
3.3.3 Choosing the right number of Principal Components
Bus used slope plus three principal components to characterize his visible data.
Because data in the extended wavelength range have many more features we use slope
plus 5 principal components. We look at the amount of variance contained within each
principal component to determine how many to use. Greater variance means more
information is contained within that principal component.
Because 40 channels were put into PCA, 40 principal components were the
output. Since PCA puts the most information in the first dimensions, and decreases with
L
O 0.0
-0.4
-0.6-------- ----------~
each successive dimension, we only want to look at some of the principal components.
To decide how many principal components we want to use in our analysis we must look
at the variance contained within each. Table 1 shows percentage variance contained
within each principal component. Note that slope is not included in this calculation, so
this percentage represents the variance left over after the slope was removed. PC10'
through PC40' contribute essentially no new information so are therefore not included in
the table. Upon inspection we see that most of the variance exists in the first six principal
components. Figure 6 shows a scree plot of the variance. Scree is rubble rock at the
bottom of a mountain. On scree plot the point where the plot levels off after the drop
determines at which principal component to stop. (Jackson 1993) We see the chart level
off around PC5' and PC6'. We decided that since 99.2% of the variance was contained
within the first 5 principal components, they were sufficient. It is evident that PC5' does
not seem to contain a significant amount of information, however, we found it useful for
classifying C-complex objects. By running PCA with the slope included in the data we
found that slope accounts for 88.4% of all the variance within the data. All the other
principal components combined account for 11.6%. Slope plus the first five PC' account
for 99.9% of the variance. Table 2 shows the variances accounted for by the slope plus
the first five principal components.
Table 1 Variance Contained within Each Principal Component (not including slope)
We find 99.2% of the variance is accounted for by using PC1' through PC5'
Principal VarianceComponent (% )PC1' 63.1PC2' 24.3PC3' 8.9PC4' 2.2PC5' 0.6PC6' 0.3PC7' 0.2PC8' 0.1PC9' 0.1PC10'-PC40' 0.1Total: 100.0
Table 2 Variance Contained within Slope and Principal Components
PC Variance (%)Slope 88.4PC1' 7.3PC2' 2.8PC3' 1.0PC4' 0.3PC5' 0.1Total: 99.9
60
40
20
0
Scree Plot
2
3
45 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40
I , , • • I , L I , , I , , , ,
0 10 20 30 40Primed Principal Component Number
Figure 6: Scree Plot. This plot shows the amount of variance contained in each principalcomponent. The further down in principal component space, the less information iscontained in each principal component. PCI' through PC5' contain most of the variancein the data set. Notice that PC5' lies just at the bottom of the downward slope indicatingit is a good place to stop using components. (Jackson 1993)
3.3.4 A Note on Notation: What is the difference between PC1' and PC1?
If the original, normalized and splined data without the slope removed were
analyzed with PCA the first transformation would be PC1. PC1 corresponds to slope.
However, we remove the slope and re-perform PCA. In this case, we denote the slope-
free PCA results as PCI', PC2', etc. We use the " ' " notation so as to recall that PC1' is
closely analogous to PC2 in the original (no slope removed) data set.
4. The Taxonomy
The most striking feature plotted from the principal components is the large gap
in PCI' versus PC2' space seen in Figures 7 and 8. It appears that this clear boundary
distinguishes between spectra that have a two micron absorption band and those that
don't. This gap is well represented by the line:
PCI' = -3.00PC2' - 0.28 (line a) (1)
All objects below and to the left of this line have no two micron absorption features and
include all featureless (C, X complex) spectra. Objects plotted to the right of and above
the line have two micron absorption features, and farther right signifies stronger
absorption. Moving right in the orthogonal direction to line ca also seems to signify
narrowing 1 micron bands. For example, V-types plot furthest to the right. The only
classes that cross this gap are the A and Sa classes. Interestingly the K-class, long
considered as an intermediate between S and C, falls most squarely in the gap. Figures 7
and 8 plot all objects in PC1' and PC2' space. Figure 7 shows the objects plotted with
original Bus labels, while Figure 8 shows the new DeMeo labels. An enlargement of the
gap present at line a is shown in Figure 9.
Ld - A
L A
LLLL RSr V
LL
A q R V
SqR
Q
0 0
1
0
-1
-0.5 0.0 0.5 1.0PC2'
Figure 7: Plot of PC1' versus PC2' for all objects. These objects are plotted with theiroriginal labels under the Bus system. Objects that do not have Bus designations arelabeled as a ".".
Sr B
AlI I u ss
_ · · _·
0
-1
-0.5 0.0 0.5 1.0 1.5PC2'
Figure 8: Plot of PC2' versus PCI'. All objects plotted are labeled with their DeMeotaxonomy classification. Notice the large gap between the S-complex and the C- and X-complexes. Line a separates objects with and without 2 [tm absorption bands. Thedirection orthogonal to line a indicates deeper 2 [tm and narrower 1 [m absorptionbands.
0.5
0-.0
-i.0
-0.4 -0.2 0.0 0.2 0.4 0.5PC2'
Figure 9: Enlargement of gap in PCI' and PC2' space. This figure shows the cleardistinction between the S-Complex and end members with 2 pm absorption bands andobjects that are featureless or have only 1 pm absorption bands.
In defining classes for the taxonomy we created a flowchart (Appendix D)
containing steps to find the right combination of principal components ranges that result
in the correct class for each object. The order of the flow chart is significant because
some classes overlap in certain principal components but can be separated in others. For
all the new labels to the 365 objects in the dataset see Appendix E. We start by
separating the A and Sa classes because they cross over the large gap in PC1' and PC2'
space. The DeMeo Sa class was relabeled from the Bus system because the two Sa
objects in this system were both Sr-types in the Bus system. Since these objects prove to
be intermediate between S and A change the name of these two Sr-types. This is step 1
in the flow chart. Refer to Appendix C for the complete chart. Figure 10 shows the
progression from S to Sa to A type.
S, Sa, A Types
2.5
2
U 1.5
4J
0.5
00.5 1 1.5 2 2.5 3
Wavelength
Figure 10: Plot of S, Sa, and A. There is a clear progression from S with a shallow onemicron band and low slope to A with a deep one micron band and high slope. Sr seem toto resemble A-type one micron band absorption very well, but are much less red than A-types.
Step two starts by separating all objects by the divide (line a) in PCI' versus
PC2' space. Step two continues to create boundaries for objects with a two micron band
and step three addresses featureless objects (the C- and X-complexes) as well as K and L
classes which have no significant two micron band.
* Sa
AS
We started by looking at the end member classes in PCA space since they
separate more clearly making them the easiest to define. In Figure 10 you can see the data
of Figure 8 with lines separating S-complex and end member classes. The V class lies to
the right of the line parallel to line a noted as equation 2:
PC1' = -3.OPC2' + 1.5 (line 8) (2)
The R class lies to the left of line 8 but right of equation 3 and above equation 4:
PC1' = -3.OPC2' + 1.0 (line y) (3)
PC1' = 1/3PC2' -0.4 (line r) (4)
To see the lines labeled in PCA space refer to Figure 11. Note that there is only one R-
type object in our sample, 349 Demboska, for which the class was created. Also, there
are only two O-type objects among our data. Even though the class is separated in the
flow chart, more data on R-type and O-type objects may change the region boundaries
significantly. To see the differences between the end member classes V, R, Q, and O see
Figure 12.
0.0 0.5 1.0PC2'
Figure 11: Plot of PC2' vs. PC1'. This plot shows the lines that divide the objects withand without 2 micron absorption bands, as well as the V and R class boundaries.
V, O, Q, R Types
0.5 1 1.5 2 2.5
Wavelength
Figure 12: Plot of V, O, Q, R Types. This plot shows typical spectra in each end memberclass. Note O has a very wide 1 micron band and V has a very narrow band. The V-typeswith the deepest 2 micron bands plot farthest from line a.
The S-complex was by far the most difficult to define. Most classes seemed to
blend together or scatter randomly in all combinations of PCA components. Sa and Si
(named under the Bus taxonomy) were entirely indistinguishable so the S1 class was
combined. Sa objects were most easily distinguishable not by absorption features, but by
their greater slope (caused by slope increases in the 1 to 1.5 micron range.) The S, Sq,
and Sk objects were initially impossible to define clearly because the boundaries seemed
to blur and overlap. We performed a new principal component analysis to help guide us.
Because the main difference between the S-classes appears to be the width of the 1
micron absorption band we used the wavelength range 0.8 to 1.35 microns and did PCA
on only S-type objects. Sa types were scattered randomly supporting that their main
difference is slope. From this new PCA (sPCA) we found that sPC1 and sPC2 best
separated the classes. We relabeled objects based on sPCA and plotted the newly changed
labels back in the original PCA.
Once we used sPCA to guide our labels, we could use PC1' and PC2' to create
definitions within our original principal components. We continued to use boundaries
parallel and perpendicular to line a. Each class has its own box in PC1' and PC2' space.
Refer to Figure 13 to see the S-complex boxes labeled in PCA space. The S class was
right of line a (equation 1), to the left of equation 5 and above equation 6.
PCI' = -3.OPC2' + 0.35 (line P) (5)
PC1' = 1/3PC2' - 0.10 (line ý) (6)
Objects that reside below line ý (equation 6) appear similar to Q-types, but with more
shallow absorption bands. These are Sq-types transitioning between S and Q. They lie
right of line a, left of line P, below line t, which is the S-type boundary, but above line rl,
which is the boundary for Q-types. Sr-types transition between S and R classes. They
reside between lines P and y, and between lines r and a given in equation 7.
PCI' = 1/3PC2' + 0.55 (line s) (7)
The objects between P and y, but above line E were unique from Sr because they
exhibited very narrow 1 lim absorption bands. These two objects appear to transition
between S and V classes. They are not included in the Bus dataset, and Bus (1999) did
not report any cases of these objects. They are given the label Sv. The Sk type was non-
unique and was absorbed into the other S classes
-0.2 0.0 0.2PC2'
Figure 13: PC2'bounding it. All
versus PC1' plotted for the S-complex. Each class has its unique boxlines are perpendicular or parallel to line a.
Sa and Sl (labeled under the Bus system) objects were indistinguishable when
extending to the near-infrared. Their features were also indistinguishable from other S-
complex objects, and it seems that their slope was the major distinction. Because the Sa
and Sl designations did not indicate that these objects were intermediate classes between
ronI
S and A or L we felt that the notation was misleading. We removed the Sa and S1
designations given by Bus and replaced them by designations of S, Sq, Sr, and Sv based
on PC1' and PC2' which represented absorption features. We felt, however, that even
though slope did not merit a class distinction, it was worth noting in the taxonomy.
The S, Sq, and Sr classes all had very widely varying slopes. High slopes are
indicative reddening by space weathering (Clark et al. 2002). Even though slope has no
mineralogical significance we thought it fitting to distinguish between objects that had
undergone significant space weathering from those who had not. We made a cutoff,
albeit an arbitrary one, at Slope = 0.25 dividing "high slope" objects from other objects.
These objects are not relabeled in a class of their own. Instead the S, Sk, and Sq objects
with high slopes get a notation of "w" added to their name to indicate a "weathered"
object. The high slope S objects are labeled Sw, Skw, and Sqw. There were also two V-
type objects and two Q-types with slopes greater than 0.25, which were labeled Vw and
Qw. To see the difference between a low and high slope object, S and Sw, refer to
Figure 14.
S, Sw
1.0
1.6
1.4
1.2
C
0.6
0.4
0.2
00 0.5 1 1.5 2 2.5 3
Wavelength
Figure 14: Plot of S and Sw. The absorption features for both are very similar. Slope isthe most significant distinction between the two.
The choice of 0.25 for the slope cutoff was somewhat arbitrary. When plotting
Bus labeled S, Sa, and S1 objects, there was a mixing around the 0.23 to 0.27 slope range.
We wanted to keep the "w" notation more selective, but also didn't want to set the cutoff
too high where objects with unusual slope features (such as deeper UV dropoffs) were
preferentially selected rather than focusing on the significant slope range between one
and two microns for the S-Complex. See Figure 15 for the plot of Slope versus PC1' that
shows the line separating weathered objects from regular objects.
* Sw
,S
·C· -------- ---
SVS V S
V Sr S S
SS S S
VSS S S S S SSr Sr SrSS S S
S S S SS S Sr SS
ss q Sq Sq S sq 5
sS S S ? S Sv S %'S S Sr S S S S S
Sq Sq Sq S S
Sq Sq SSq SqQ Sq Sq Sq
Q Sq
00
0.2
slope
Sw
Srw
SW S Vw OwSw Sw
Sw~ SwSrw Sw •w
SSw 5'Sw
Sqw Sqw Sqw Sq~ u
Cw SqwSqw R Sqw
Sqwow
Figure 15: Plot of Slope versus PCI' for the S-Complex. All objects with slopes greaterthan 0.25, regardless of their subclass (S, Sq, Sk), get a "w" notation indicatingweathering. Note the two V and Q class objects labeled "Vw" and "Qw." Dembowska, asthe only R member, does not get a w.
Step three focused on objects below or to the left of line a (equation 1). We again
start by removing end members. D and T classes are easily separated by their high slopes
and PCI', PC2', and PC3' values. Next, we separate out L objects based on the PC2'
versus PCi' plot. When an object fits in the L component space it is necessary to check
for Xe type objects. Xe is a class defined in the Bus system that has a distinct hook at
0.49 microns. By visually inspecting the spectrum, one can notice the presence of a
Sw
Sw SW
Sqw
Qw
Sw
SW
Sa
I ' ' 1
, I ,
feature at 0.49 and an absence of a slight feature around 1 micron and label the object an
Xe instead of an L. Refer to Figure 16 to see Slope versus PC1' which shows how D, L,
and L are fairly distinct. The K types can then be distinguished clearly in PC2' and PC3'
space. We found that all Ld type objects under the Bus system could be fit in either the L
or D classes when near-infrared data were added. Since it is not necessary for
distinguishing classes, it was removed from the DeMeo system. Figure 17 shows how
the K class is distinct.
HIope
Figure 16: Plot of Slope v. PC1'. Note how D, T, and L are fairly distinct in thiscomponent space. K separates better in PC2' and PC3' space.
LL L L LLL D
L L
Xe L L D
L Xae L T D
Xe K K Xk T T 0
K XK Xe K T
K Xc Xk XkCKKK Xk
B Ch ChhChXe Xk Ch x 08 Ch Xk 0
Cgh Ch KCbS CC CCgh Ch C (f#ChCh Cb C K X
Ch Ch CbCCh C C
C CCC
I I I I I R I I I I I
0.0
-0.2
-0.4
-0.2 -0.1 0.0 0.1PC3'
Figure 17: Plot of PC3' v. PC2'. The K types are distinct in this component space.
This brings us to the X and C complexes. The B-types are easily distinguished by
their negative slope as well as negative PC1' and PC4' values. The X-class can then be
easily removed based on high slope values between 0.2 and 0.425. At this point Xk
objects may be in X-class PC' space. When visually inspecting the spectrum if there is a
feature present at 1 micron, the object is designated an Xk. Figure 18 shows a plot of the
C- and X- complexes.
C Ch C C Cgh KC CCh C Cg
Cb X Xk C KC ChCh CCh C C X ChcgB xcg Xk KC K K K
Xk KX K LKK K K
D Xk Xk Xk KS0 X e XL
T D rr Xe LT Xe
T T Ye LL L L0 D Xe L
LD L L L
L L LL L
LLL
LL
L
-0.3 -0.2 -0.1 0.0PC2'
Figure 18: Plot of PC2' v. PCI' for the C- and X-Complex. These components helpseparate the two complexes. Slope will also differentiate the two.
The Ch objects are defined well in PC1' and PC4' space, but a check must be
done to see if a strong 0.7 micron feature exists, making the object a Cgh. The C and Cb
types separate in PCi', PC4', and PC5'. See Figure 19 and 20 to see the C-complex
plotted in PCA space. After this the Cgh, Ch, Xk, Xc, and Xe types don't all separate
cleanly in component space and are heavily dependent on visually detecting features.
0.05
0.00
-0.05
SI I j I Cg I I I I I
Cg
Cgh
Cgh
Cgh ChCgh Ch
Cg Cgh ChCh
C CýhCh CfChCh Ch ChChCh ~ Ith Ch
C C ChC
CC
C CCb
-C CbCbCb
I , , , , I l , i I , l , , I l
-0.1 0.0 0.1 O.ZPC4'
Figure 19: Plot of PC4' v. PC5' for C-Complex. The classes are easily distinguishable inthis component space.
Cgh
Ch
ChCh Ch
Ch Ch
ChCh
ChCgh
Ch
Ch
Ch
Ch
CghCh
Ch ChC
Ch C
Ch
0.00PCi'
Cg
Cgh
0.05
Figure 20: Plottypes.
of PC5' v. PC1'. A second plot demonstrating the separation among C-
We currently do not have a method for classifying spectra using only near-infrared
data because putting reflectance values into the eigenvectors from this work would be
meaningless since the entire range doesn't exist. In future work we hope to include a
method for using near-infrared only data to find a taxonomic classification. However
-0.4
-0.5
C Cb
Cb
C C
-0.05
1
I I I I·_ ~
m m I I I I I I I I
I I
, I , , ,I I
note: certain classes evolve unchanged from the Bus taxonomy that are based on features
measured at visible wavelengths. Assignment to these classes (Cg, Cgh, Xc, Xe, Xk)
therefore requires visible wavelength data.
5. Conclusion
A new taxonomy with was created using Principal Component Analysis and
visible features to characterize visible and near-infrared wavelength spectral reflectance
values. The system, based on the Bus visible taxonomy from Bus (1999), has 24 classes
compared to 26 in the Bus system. The changes in classes are summarized in Appendix
A. We used all of the seven possible changes from the previous system to the new
highlighted in Figure 3. We eliminated three classes: Ld, S1, and Sk. All the S subclasses
(Sa, S1, Sk, Sq, Sr) had objects that merged back into the S class, although many Sq
objects remained Sq and two Sr objects were relabeled Sa. A new intermediate class, the
Sv class, was created to branch the S and V classes. High-sloped S, Sa, Sq, Sr, V and Q
objects were given a "w" notation to indicate weathering. Many of the classes that lie left
of line a are either featureless or exhibit only small features at visible wavelengths
identified by Bus (1999). It is still necessary to use these visible features to distinguish
the classes because there are no other distinguishing features at near-infrared
wavelengths. 365 objects were given types based on the DeMeo taxonomic system which
was created using 6 dimensions including Slope and PCI' through PC5' of Principal
Component Analysis.
References
Barucci, M.A, Capria, M.T., Coradini, A., Fulchignoni, M. (1987). Classification of asteroids
using G-mode analysis. Icarus 72, 304-324.
Binzel, R.P., Xu, S. (1993). Chips off of asteroid 4 Vesta: Evidence for the parent body of
basaltic achondrite meteorites. Science 260, 186-191.
Binzel, R.P., Rivkin, A.S., Stuart, J.S., Harris, A.W., Bus, S.J., Burbine, T.H., (2004). Observed
spectral properties of near-Earth objects: results for population distribution, source
regions, and space weathering processes. Icarus 170, 259-294.
Bowell, E., Chapman, C.R., Gradie, J.C., Morrison, D., Zellner, B. (1978). Taxonomy of
Asteroids. Icarus 35, 313-335.
Bus, S.J., (1999). Compositional structure in the asteroid belt: results of a spectroscopic survey.
Doctoral thesis. Massachusetts Institute of Technology, Cambridge, Massachusetts.
Bus, S.J., Binzel, R.P., (2002). Phase II of the Small Main-Belt Asteroid Spectroscopic Survey.
Icarus 152, 106-145.
Chapman, C.R., Johnson, T.V. (1971). A review of spectrophotometric study of asteroids.
Physical Studies of Minor Planets (T. Gehrels, Ed.), 51-65. NASA SP-267.
Chapman, C.R., Morrison, D., Zellner, B., (1975). Surface properties of asteroids: A synthesis of
polarimetry, radiometry, and spectrophotometry. Icarus 25, 104-130.
Clark, B.E., Hapke, B., Pieters, C., Britt, D. (2002) Asteroid Space Weathering and Regolith
Evolution. In Asteroids III, (W.F. Bottke, A. Cellino, P. Paolicchi, R.P. Binzel eds.),
Univ. Arizona Press, 585-599.
Consulmagno, G.J., Drake, M.J. (1977). Composition and evolution of the eucrite parent body -
Evidence from rare earth elements. Geochimica et cosmochimica Acta. Vol 41 Isssue 9,
1271-1282
DeMeo, F.E., Binzel R.P (2007). Comets in the near-Earth Object Population. Icarus. Submitted
for publication.
Gaffey, M.J., Bell, J.F., Brown, J.R., Burbine, T.H, Platek, J.L., Reed, K.L., Chaky, D.A., (1993).
Mineralogic Variations within the S-Type Asteroid Class. Icarus 106, 573-602.
Jackson, D.A. (1993) Stopping rules in Principal Component Analysis: A comparison of
heuristical and statistical approaches. Ecology 74 2202-2214.
Rayner, J.T., Toomey, D.W., Onaka, P.M., Denault, A.J., Stahlberger, W.E., Vacca, W.E.,
Cushing, M.C., Wang, S. (2003) SpeX: A medium-resolution 0.8-5.5 micron
spectrograph and imager for the NASA Infrared Telescope Facility. Astron. Soc. of the
Pacific 115 362-382.
Tholen, D.J. (1984) Asteroid taxonomy from cluster analysis ofphotometry. Doctoral Thesis,
University of Arizona.
Tholen, D.J., Barucci, M.A. (1989) Asteroid taxonomy. In Asteroids II (R.P.Binzel, T.Gehrels,
and M.S. Matthews, Eds.), 298-315. Univ. of Arizona Press.
Tody, D. (1993). IRAF in the Nineties. In Astronomical Data Analysis Software and Systems II
(R. J. V. Brissenden, and J. Barnes, Eds.), pp 173. Astron. Soc. of the Pacific, San
Francisco.
Wood, J.H. Kuiper, G.P (1963) Photometric studies of asteroids. Astrophys. J. 137, 1279-1285.
Vernazza, P., Birlan, M., Rossi, A., Dotto, E., Nesvorny, D., Brunetto, R., Fornasier, S.,
Fulchignoni, M., Renner, S. (2006) Physical characterization of the Karin Family. AA
460, 945-951
Appendix A: Bus-DeMeo class changes
DeMeo
B PICCb ,1CgCgh .•.Ch a-D -
Ld-L 'cK .0O ---- *Q -R
SqSr
---- V
-*---
SaS1 -,Sk k
T --T
BCCb
CgCghChD
LK0Q,QwRSr, SrwSq,SqwSa
S,SwSvT
V ----- V,VwX - XXc - XcXk----- XkXe - Xe
Total: 26 24
Eliminated: Created:Ld, Sk, SI Sv
w notation does not denote a distinct class.
Bus
Appendix B: Observation List
Object12345781011131415161718192021222425262728293032333437383940414243484950515255565758616364
Obs Date19-May-0529-Mar-0117-Mar-039-Oct-0020-Feb-0420-Feb-041 6-Sep-0219-Feb-0413-Nov-0519-May-0517-May-0117-Aug-0221-Jan-0715-Aug-0122-Jun-0129-Jan-0622-Jun-0122-Sep-0425-Oct-068-Oct-0530-Jan-0124-Aug-011-Jun-0213-Jan-0229-Jan-018-Oct-0030-Jan-016-Mar-0220-Feb-041 5-Aug-01
5-Jul-0314-Aug-0116-Oct-0415-Sep-0421-Jun-0122-Nov-058-Oct-052-Aug-035-Jul-03
1 5-Jun-0428-Jun-0629-Jan-0622-Sep-041-Jun-022-Aug-038-Oct-00
30-Sep-0330-Jan-01
TelescopeIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3m
KPNO 4mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3m
Object65666769707376777882848587909293969799101103105106108110111114115119128130131132133147150151153158160170175180181188191192199
Obs Date30-Sep-0322-Nov-0516-Jun-0411-May-0529-Sep-0216-Oct-035-Sep-0525-Oct-0631-Oct-0524-Aug-012-Aug-032-Aug-034-Sep-055-Sep-058-Oct-00
27-Apr-0328-Jan-068-Oct-05
13-Nov-0522-Dec-0614-Aug-012-Aug-032-Aug-03
31-May-0229-Jan-0122-Sep-0428-Jun-066-Mar-0221-Jun-012-Aug-0329-Mar-0120-Jul-065-Sep-0522-Jun-012-Aug-032-Aug-0313-Nov-0510-May-0516-Mar-035-Jul-03
22-Jun-0113-Nov-0528-Sep-0230-Sep-031 5-Apr-0212-Apr-0530-Apr-0617-Mar-03
TelescopeIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3m
Object Obs Date Telescope Object Obs Date Telescope201 19-Feb-04 IRTF 3m 596 1-Jun-02 IRTF 3m205 2-Aug-03 IRTF 3m 599 19-Feb-01 IRTF 3m210 13-Nov-05 IRTF 3m 606 29-Sep-02 IRTF 3m221 8-Oct-00 IRTF 3m 625 20-Feb-01 IRTF 3m236 01-May-06 IRTF 3m 631 29-Sep-02 IRTF 3m237 15-Apr-02 IRTF 3m 653 17-Mar-03 IRTF 3m243 1 7-Mar-03 IRTF 3m 661 1 6-Mar-03 IRTF 3m244 29-Mar-01 IRTF 3m 670 29-Sep-02 IRTF 3m246 9-Mar-05 IRTF 3m 673 21-Jun-01 IRTF 3m250 1 6-Oct-03 IRTF 3m 675 15-Apr-02 IRTF 3m258 24-Aug-01 IRTF 3m 679 19-Feb-01 IRTF 3m264 19-May-05 IRTF 3m 688 28-Sep-02 IRTF 3m266 5-Sep-05 IRTF 3m 699 8-Jan-05 IRTF 3m267 2-Aug-03 IRTF 3m 706 30-Jan-06 IRTF 3m269 11-May-05 IRTF 3m 716 24-Aug-01 IRTF 3m278 31-May-02 IRTF 3m 719 23-Aug-01 IRTF 3m279 8-Jan-05 IRTF 3m 720 9-Oct-00 IRTF 3m288 29-Sep-02 IRTF 3m 729 14-Aug-01 IRTF 3m289 29-Jan-01 IRTF 3m 739 28-Sep-02 IRTF 3m295 28-Sep-02 IRTF 3m 742 1 6-Mar-03 IRTF 3m308 18-Apr-05 IRTF 3m 773 22-Sep-04 IRTF 3m322 29-Sep-02 IRTF 3m 776 1 2-Apr-05 IRTF 3m337 28-Sep-02 IRTF 3m 782 22-Jun-01 IRTF 3m345 8-Oct-05 IRTF 3m 785 11-May-05 IRTF 3m346 31-May-02 IRTF 3m 789 17-Aug-02 IRTF 3m349 22-Jun-01 IRTF 3m 793 28-Sep-02 IRTF 3m352 22-Jun-01 IRTF 3m 808 14-Aug-01 IRTF 3m354 15-Apr-02 IRTF 3m 824 14-Aug-01 IRTF 3m359 30-Jan-01 IRTF 3m 832 5-Jul-03 IRTF 3m371 14-Aug-01 IRTF 3m 832* 3-Mar-05 CFHT 3.6m378 14-Apr-02 IRTF 3m 847 19-Feb-01 IRTF 3m387 27-Apr-03 IRTF 3m 863 14-Jan-02 IRTF 3m389 24-Aug-01 IRTF 3m 908 20-Jul-06 IRTF 3m402 19-Feb-01 IRTF 3m 913 15-Aug-01 IRTF 3m403 27-Apr-03 IRTF 3m 925 28-Sep-02 IRTF 3m433 17-Aug-02 IRTF 3m 944 22-Sep-04 IRTF 3m434 23-Aug-01 IRTF 3m 944 14-Jun-01 Magellan 6.5m444 1 5-Aug-01 IRTF 3m 984 29-Sep-02 IRTF 3m446 14-Aug-01 IRTF 3m 985 11-May-05 IRTF 3m453 1 5-Aug-01 IRTF 3m 1011 1 2-Jan-02 IRTF 3m456 17-Jun-02 IRTF 3m 1020 17-Mar-03 IRTF 3m460 17-Jun-02 IRTF 3m 1036 28-Mar-01 IRTF 3m485 6-Mar-02 IRTF 3m 1036 26-Jan-04 IRTF 3m512 16-Oct-04 IRTF 3m 1036 9-Mar-05 IRTF 3m513 29-Mar-01 IRTF 3m 1036 03-Jun-06 IRTF 3m532 21-Jun-01 IRTF 3m 1065 8-Mar-05 IRTF 3m570 20-Feb-04 IRTF 3m 1094 1 6-Mar-03 IRTF 3m579 20-Feb-01 IRTF 3m 1126 20-Feb-01 IRTF 3m
*Data from Vernazza et al. (2006)
Object113111311139113911431147114811981204122813001329133213501374143314591471149415081542156516201640164216581659166016621667168517511807183918481858186218621864186618661903190419161929194319431980
Obs Date19-Feb-0411-May-0516-Jun-045-Sep-0510-Jun-0515-Aug-0129-Mar-0128-Oct-0219-Feb-041 6-Mar-0325-Oct-0617-Aug-0210-Jun-0521-Jun-0120-Feb-0417-Aug-0220-Feb-018-Oct-05
25-Oct-0619-Feb-0418-Apr-053-Mar-05
29-Jan-018-Mar-05
15-Aug-0130-Jan-011 5-Mar-028-Mar-0517-Mar-0328-Oct-029-Mar-0522-Jun-011 5-Sep-0429-Mar-0115-Aug-0114-Apr-0213-Nov-0522-Nov-0529-Mar-016-Jan-94
21-Nov-0617-Mar-034-Sep-00
14-Aug-0119-Feb-0112-Jan-026-Nov-9925-Oct-06
TelescopeIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3m
MDM 2.4mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3m
Palomar 5mIRTF 3m
Object203520422045206320642074208520992107215722462335235323542378238623962401244224482501250425212566257927152732285128732875291129122957296529773028310231033122315531983199320032483255331733633395
Obs Date28-Oct-0223-Aug01?14-Jan-0219-May-0529-Jun-0630-Sep-0314-Apr-028-Oct-0529-Jan-061 6-Mar-0317-Apr-0530-Oct-0531-May-0217-Mar-0311-May-0517-Jun-0215-Aug-0117-Aug-0215-Sep-021 6-Mar-0312-Jan-0227-Apr-031 3-Jan-0216-Sep-0210-Oct-001 5-Aug-011 5-Aug-0124-Aug-0113-Jan-021 6-Mar-0217-Mar-0320-Feb-011 6-Mar-0310-May-0528-Sep-0216-Sep-029-Oct-0021-Jun-0126-Jan-0422-Jun-0110-May-053-Mar-0510-Dec-041 8-Apr-0516-Sep-0201-May-0615-Apr-021 9-Feb-01
TelescopeIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3m
Object340234023430349135113628363537013734375337883844385838733903390839103920394940384055414241794188419743524407441744514558457046884713473749955013511151435230526152615401540755875604560456415660
~Obs Date30-Sep-031-Sep-0317-Jun-0217-Jun-0215-Apr-0230-Apr-0612-Nov-0514-Aug-0124-Aug-018-Oct-0517-Mar-0327-Apr-0320-Feb-0428-Oct-021 6-Sep-021 5-Sep-0417-Aug-0216-Sep-0224-Aug-0128-Oct-0211-Apr-0515-Apr-0215-Sep-0414-Aug-0112-Oct-9617-Jun-0224-Aug-0124-Aug-0114-Jan-0215-Jun-0421-Jun-0129-Jan-0114-Apr-0217-Jun-0228-Oct-0214-Jan-025-Sep-05
25-Oct-065-Sep-05
11-May-0519-May-051 6-Mar-0313-Jan-0228-Mar-0129-Mar-011-Mar-0012-Apr-0522-Aua-93
TelescopeIRTF 3m
KPNO 4mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3m
MDM 2.4mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3m
KPNO 4mIRTF 3m
MDM 1.3m
Object56855817584060476239638664116455658573417763833484441727418736191271935619356207862079022771244753510736284534355343554690546906614686450868198935598943989431370621989 VA1997 AE122000 GD22000 PG32000 XL442001 MQ32001 TX162001 XN2542002 AA2002 AV
Obs Date29-Mar-0122-Sep-0424-Aug-018-Oct-0526-Jan-043-Mar-05
20-Feb-0428-Oct-0225-Oct-0616-Mar-021-Jun-02
17-Mar-0330-Sep-0310-Oct-0030-Jan-0130-Sep-0330-Jan-0110-Apr-971-Sep-03
29-Jan-0116-Oct-0329-Mar-0127-Dec-0216-Mar-0225-Oct-0612-Jan-0028-Mar-0117-May-011-Sep-0329-Jan-019-Oct-0016-Mar-0224-Oct-0424-Dec-0127-Oct-0227-Oct-0216-Oct-0315-Mar-024-Sep-0029-Jan-0114-Aug-011 5-Mar-0214-Apr-0213-Jan-0213-Jan-02
TelescopeIRTF 3mIRTF 3mIRTF 3mIRTF 3m
KPNO 4mIRTF 3m
KPNO 4mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3m
MDM 2.4mKPNO 4mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3m
Palomar 5mIRTF 3m
KPNO 4mKPNO 4mIRTF 3mIRTF 3mIRTF 3mIRTF 3m
Palomar 5mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3mIRTF 3m
Y
Appendix C: Principal Component Eigenvectors
Wavelength Eigen Eigen Eigen Eigen Eigen(microns) PC1' PC2' PC3' PC4' PC5'
0.45 -0.0766 -0.0643 -0.2724 0.3046 -0.51740.50 -0.0391 -0.0279 -0.1270 0.1525 -0.18760.60 0.0438 0.0176 0.1128 -0.1486 0.05930.65 0.0876 0.0343 0.2104 -0.2677 0.07540.70 0.1256 0.0471 0.2726 -0.3386 0.05230.75 0.1466 0.0096 0.2475 -0.3284 -0.02310.80 0.1271 -0.1186 0.1486 -0.2392 -0.14660.85 0.0888 -0.2673 0.0420 -0.1453 -0.25690.90 0.0680 -0.3645 -0.0385 -0.0921 -0.22930.95 0.0857 -0.3743 -0.1168 -0.0505 -0.06571.00 0.1371 -0.2899 -0.2083 -0.0289 0.10771.05 0.1921 -0.1527 -0.2809 -0.0277 0.17171.10 0.2322 -0.0381 -0.2747 -0.0160 0.16851.15 0.2566 0.0306 -0.2169 0.0077 0.16111.20 0.2704 0.0708 -0.1713 0.0304 0.14631.25 0.2787 0.1053 -0.1427 0.0450 0.10611.30 0.2849 0.1385 -0.1031 0.0608 0.05331.35 0.2852 0.1598 -0.0407 0.0842 0.00901.40 0.2782 0.1645 0.0243 0.1104 -0.04291.45 0.2641 0.1520 0.0930 0.1387 -0.08681.50 0.2427 0.1192 0.1562 0.1609 -0.11881.55 0.2154 0.0689 0.2021 0.1752 -0.12501.60 0.1841 0.0089 0.2231 0.1804 -0.11581.65 0.1531 -0.0514 0.2215 0.1714 -0.09401.70 0.1247 -0.1069 0.2043 0.1550 -0.07571.75 0.1002 -0.1532 0.1784 0.1421 -0.05251.80 0.0804 -0.1884 0.1508 0.1279 -0.02711.85 0.0665 -0.2136 0.1225 0.1095 0.01041.90 0.0570 -0.2283 0.0923 0.0868 0.04731.95 0.0513 -0.2317 0.0617 0.0610 0.07852.00 0.0502 -0.2233 0.0346 0.0358 0.10502.05 0.0538 -0.2023 0.0136 0.0103 0.12492.10 0.0607 -0.1706 -0.0038 -0.0162 0.12412.15 0.0690 -0.1302 -0.0229 -0.0476 0.09162.20 0.0778 -0.0852 -0.0447 -0.0838 0.03542.25 0.0859 -0.0406 -0.0678 -0.1225 -0.03272.30 0.0934 0.0023 -0.0911 -0.1644 -0.11262.35 0.0997 0.0438 -0.1153 -0.2068 -0.19932.40 0.1050 0.0832 -0.1389 -0.2445 -0.28842.45 0.1090 0.1177 -0.1580 -0.2708 -0.3767
Appendix D: Flow chart. It's as easy as 1, 2, 3...
Step 1
Does it satisfy:PC > -0.44
no
Does it satisfy:0.55 < Slope < 1.5
no
Does it satisfy:0.44 < Slope < 0.55
I no
Indeterminate
yes StepStep 2
yes
yes ,
A-Type
Sa-Type
SStep 2
Where does it lie compared 1to line (a? below
above
Does it lie above line 6? yes
no
Does it satisfy both: yes0.04 < Slope < 0.07-0.8 < PCl < -0.75
no
Does it satisfy both:Lies below rl
0.07 < Slope < 0.44
no
Does it satisfy both:Lies between lines y and 6
Lies above line r1
V-Type - Slope > 0.25 yes+ Add w notationVw
O-Type
yes d if yesy-e- • Q-Type •-* Slope > 0.25 - Add w notation
Qw
yes R-Type
no
S Complex
IStep 3 !
EStep3 IDoes it satisfy both:0.425< Slope < 1.25
-0.44 < PC1 < 0.4
Does it satisfy all:0.25 < Slope < 0.5
-0.28 < PC2 < -0.20-0.2 < PC3 < -0.12
no
Does it satisfy both:0.07 < PC1 < 1.00-0.5 < PC2 < -0.15
no
Does it satisfy both:-0.075 < PC3< 0.14
-0.2 < PC2 < 0.1
no
yes
yes
yes---- +
yes-------- +
D-Type
T-Type
L-Type Check for Xe:feature at 0.49 ptm.
K-Type
IC and X ComplexesMMMWMý
and X Complexes
Does it satisfy all:-0.2 < Slope < 0.0
-1.2 < PCI < 0PC4 < 0.0
no
Does it satisfy:0.2< Slope < 0.425
no
Does it satisfy both:0.01 < PC4 < 0.14
-0.75 < PC < -0.27
no
Does it satisfy both:-0.04 < PC4 < 0.02-0.07 < PC5 < -0.04
no
Does it satisfy both:-0.85 < PC1 < -0.45-0.06 < PC5 < 0.02
no
Does it satisfy both:0.02 < PC5 < 0.1-0.6 < PCI < -0.1
Does it satisfy both:-0.45 < PC I< 0. 1
-0.06 < PC2 < 0.03
no
Does it satisfy both:-0.1 < PCI< 0.3
-0.3 < PC2 < -0.2
no
Indeterminate
B-Type
X-Type
yes
yes-- =---
yesi--+
yes----- -
yes---y
Ch-Type
Check for Xk:Feature at 1 prm.
Check for Cgh:0.7 ttm featureand UV dropoff
Cb-Type
C-Type
Cgh, Cg
Xk, Xc-Type
Xe-Type
Check for Cg:no 0.7 tpm feature
Xk has 1 gm feature.Xc is featureless.Check for Xe:feature at 0.49 ,tm.
Check for Xe:feature at 0.49 rtm.
IC 0
S Complex I
Does it satisfy both:Lies left of f3 and above t.
noDoes it satisfy all:
Lies between j3 and yand between Ei and 1.
no
Does it satisfy all:Lies between p andand between E and i1.
no
Does it satisfy both:Lies between 13 and
y and above F.
·no
Indeterminate
S-Type and if Slope > 0.25 y
yst Sq-Type an Slope > 0.25
yesy--
yes
Sr-Typeand if yesSr-Tye N, Slope > 0.25
Add w notationSw
Add w notationSqw
Add w notationSrw
Sv-Type Slope > 0 yes Add w notationSvw
Checks* for Cg, Cgh, Xc, Xe, XkCg: Strong UV absorption feature before 0.55 atm
Cgh: A Cg with a broad, shallow absorption feature at 0.7[tm.
Xc: Red and featureless with slight concave-down curvature.
Xe: Concave-up absorption feature before 0.55 Vm.
Xk: Red with a flat section around 0.75 ,m.
*These spectral features all exist in the visible and wereidentified in the Bus system (Bus 1999)
EquationsPC1' = -3PC2' - 0.28 Line a
PC1' = -3PC2' + 0.35 Line p3PC1' = -3PC2' + 1.00 Line y
PC1' = -3PC2' + 1.50 Line 8
PC1' = 1/3PC2' + 0.55 Line E
PC1' = 1/3PC2' - 0.10 Line t
PC1' = 1/3PC2' - 0.40 Line r1
m
Bus Class
C
B
Sk
V
SS
~
Xe 0.1201 0.0463 -0.2407 -0.0967 -0.0392 -0.0059
Appendix E: New Labels to Data
~Object
12
3
4
5
7
810
11
13
1415
16
17
18
19
2021
22
24
25
26
2728
29
30
32
33
34
37
38
39
4041
42
43
48
49
5051
52
55
56
57
58616364 Xe
DeMeo Class
C
B
SqV
S
SqSw
C
SqCh
S
SqX
S
S
Ch
SXk
X
C
S
S
S
S
SS
Sw
SCh
S
CghSqw
S
SqK
SqCh
Ch
Ch
Ch
C
XXk
S
Ch
SS
Slope0.0060
-0.08920.1775
0.02700.1074
0.1930
0.29020.1391
0.21270.07570.1706
0.1581
0.28100.2103
0.1636
0.1935
0.08110.06780.30080.05520.20620.1836
0.22680.13780.20080.2165
0.2623
0.09960.07170.1782
0.12190.27200.1209
0.1834
0.21320.1972
0.0207
0.00780.04600.1844
0.1779
0.26680.1499
0.1680
0.01670.16640.2382
PC1'
-0.5626-0.5691-0.1930-0.00430.1467
-0.11410.2597
-0.7550-0.3055-0.41390.0743
-0.1553-0.02530.4067
0.2243-0.56540.0203
-0.0574-0.3406-0.83100.1640
-0.0214
0.07120.29680.12640.3434
0.08780.0297
-0.33250.2766
-0.2905-0.13720.0628
-0.3458-0.0639-0.1731-0.6803-0.4755-0.5934-0.3110-0.6801-0.3587-0.14230.2697
-0.46200.34110.3774
PC2'
-0.0669-0.06400.0610
0.6020-0.0250-0.0240-0.13990.02460.0236-0.0506-0.0488-0.0189-0.1749-0.1167-0.1079-0.01020.0044
-0.1942-0.13020.0332
-0.01070.0007
-0.0064-0.0959-0.1190-0.16210.03230.0872
-0.0385-0.0926-0.0760
-0.0209-0.09450.0588
-0.16950.08160.0187
-0.00950.0395
-0.0788-0.0223-0.0564-0.1582-0.0700-0.0242-0.0905-0.1165
PC3'
-0.1351-0.1550-0.0033-0.1556-0.01610.06540.1139
-0.1519-0.0003
-0.1363-0.03630.0614
-0.1528-0.0095-0.0199
-0.1606-0.0221-0.1419-0.1396-0.21650.1429
0.11350.12120.0218
-0.01600.0467
-0.0068
-0.0577
-0.18900.0083
-0.14580.1027
0.0195-0.11600.11330.0944
-0.1861-0.1750-0.2039-0.0922-0.1447-0.1573-0.1015
-0.0333
-0.1881-0.00710.1057
PC4'
-0.0871-0.09160.0218
0.0189-0.0212-0.0415-0.0320-0.0311
-0.01680.08490.0153
-0.0094
0.00540.0088
-0.0013
0.0874-0.04110.0333
-0.0003
-0.09560.01440.01760.0233
-0.01490.0138
-0.0064
0.04170.05510.0935
-0.01770.06870.0434
-0.06150.1253
-0.04710.06600.05560.0830
0.09270.1159
-0.11420.0262
-0.05690.01980.0638
-0.0124-0.0713
PC5'
-0.0256-0.04580.0038
-0.0255
0.0158-0.0042-0.0538-0.0107
-0.00160.00800.0095
-0.0116-0.00130.0096
0.01950.0176
0.0071-0.03170.0118
-0.0358-0.02790.0070
-0.00260.00900.0041
-0.0149
0.03300.03530.00070.01270.0450
-0.0457
-0.0012-0.0192-0.0477-0.0134
0.00050.02030.00790.04680.0066
-0.00150.00140.0065
0.01080.0126
-0.0184
C
Sk
Ch
S
SX
SI
SCh
SXk
X
B
S
S
S
SS
SI
S
SqCh
S
CghS
SCh
L
Sk
Ch
Ch
Ch
Ch
C
XXk
SCh
SSa
Object Bus Class DeMeo Class Slope PC1' PC2' PC3' PC4' PC5'
65
66
67
6970
73
76
77
78
82
84
85
87
90
92
9396
97
99
101103105
106108110111
114115119128130131
132133147150151153158160170175180181188191192199
X
Xe
Ch
Sq
ChB
X
C
Xc
CT
X
Xk
S
S
Ch
CghSIX
Ch
Xk
SSI
CCh
Xc
Xe
S
C
Cb
SI
X
S
CS
CgSqXk
S
Cb
SIX
Xk
Ch
SX
Ch
S
X
Xe
Ch
S
Ch
C
X
C
Xk
C
TXc
Xk
S
SCh
CghSw
XCh
K
S
SC
Ch
K
Xe
S
C
CSw
X
S
Ch
S
CgSr
Xk
S
CbSw
X
0.1823
0.1009
0.15340.29460.12140.21820.24790.1781
0.11920.05140.07640.10470.2873
0.1902
0.1604
0.02350.26720.0782
0.09060.1685
0.1399
0.09010.04600.27130.2413
-0.00380.22820.1408
0.23290.0909
0.09250.1741
0.1041
0.17150.1778
0.15580.37800.25340.20670.1421
0.1472
0.11760.11610.03170.1690
0.1738
0.30860.4045
-0.4111
-0.47390.1438
-0.3394-0.27710.4242
-0.6309-0.0892
-0.50450.0305
-0.4183
-0.7800-0.3383-0.7534
-0.2105-0.56090.0024
-0.1431
-0.09860.1674
0.1377
-0.3221-0.50450.2950
-0.2178-0.3827-0.2364
0.33460.2297
-0.4973
-0.5175-0.5109-0.36400.1898
-0.6347-0.66580.3427
-0.4440-0.0603-0.38800.2552
-0.55370.1465
-0.04030.3672
-0.55790.2612
-0.1176
-0.0376-0.03770.0272
-0.0517-0.0617-0.0500-0.0148-0.2235-0.03020.0415
-0.04800.0348
-0.08660.0431
-0.0825-0.0721-0.2296-0.2083-0.1395-0.0220-0.0870-0.07850.0048
-0.0596
-0.0643-0.0731
-0.1662-0.0679-0.0735-0.08270.0020
0.0313-0.0944-0.0812-0.0061
-0.0221
-0.1232-0.05510.0573
-0.0567-0.0862-0.01300.0934
-0.1862-0.0820-0.0200-0.0848-0.1249
-0.1855-0.17990.0028
-0.1497-0.1589-0.0310
-0.1679-0.1153-0.1790
-0.0687-0.1660-0.1901-0.1944-0.1747-0.1616-0.1823
-0.1353-0.1168
-0.11460.03470.0598
-0.1738-0.12150.0671
-0.1567-0.1973
-0.0634
0.02580.1495
-0.1416
-0.1308
-0.0507
-0.1157
0.1374-0.1881-0.19460.1094
-0.17020.0102-0.15720.0138
-0.1130-0.0583-0.0815
0.0267-0.19740.1100
-0.1555
-0.04640.0362
-0.01780.02160.11650.0453
-0.0355
-0.02440.03720.04790.0696
-0.0317
-0.0119
-0.0556
0.0231-0.12020.0032
-0.0371
0.0174
-0.0371
-0.09320.06290.0594
-0.00200.08370.0372
-0.0363
-0.0233
-0.0176-0.0643
0.0706
-0.0184-0.0160
-0.0473
-0.0282-0.0545
0.06870.01720.00800.0260
-0.0673-0.05580.0595
-0.03890.0087
-0.00770.00200.0677
-0.01150.00710.02140.00600.03090.0243
-0.0054-0.01410.0241
-0.0166-0.0023-0.0360-0.0171-0.0195
-0.0113-0.0490-0.02210.00350.01490.0377
-0.0041-0.00170.05840.0111
0.0076-0.0125-0.02610.0095
-0.0105-0.01330.0491
0.0174
0.0129-0.0146-0.0113-0.02960.0251
-0.00730.03490.00520.02080.02880.0310
-0.03900.0322
-0.0406-0.0166-0.0176
Object Bus Class DeMeo Class Slope PC1' PC2' PC3' PC4' PC5'201205210221236237243244246
250258264266267269278279288289295308322337345346349352354359371378387389402403433434444446453456460485512513532570579
XChCb
K
LSr
SSw
AXk
SS
ChDDS
DSA
SwT
XXkCh
SR
SwA
XkS
SLSL
SSwXe
CA
Sw
SL
S
SqwK
ST
K
0.20040.05800.09040.01580.21440.18000.21910.35610.60360.25860.13880.20220.09880.4505
1.24780.24410.58860.16421.38270.29250.31870.40130.1473
0.12860.17860.32130.29490.73420.11630.17210.16370.11440.11840.04710.1814
0.27620.09190.1228
0.78090.25470.2361
-0.00500.1651
0.29180.01570.17810.41690.1296
-0.2912-0.3641-0.6191-0.25980.21800.37350.04100.3378
-0.5975-0.13080.24120.1715
-0.56080.26690.10620.1226
-0.42330.4438
-1.66200.04530.0562
-0.3421-0.3521-0.7064-0.0587-0.2178-0.0003-0.4539-0.25900.38120.36100.53200.08290.1322
-0.00470.07870.0091
-0.6121-0.7220-0.03380.11840.43650.1581
-0.1499
-0.06490.01970.1411
-0.1851
-0.0816-0.0512-0.0086-0.1234-0.25860.01360.0275
-0.1351-0.0238-0.09050.0206
-0.00410.0061
-0.2875-0.22370.0262
-0.0246-0.12520.3082
-0.0111-0.2577-0.0455
-0.03200.02980.05210.4352
-0.00870.0124
-0.0821-0.0676-0.1108-0.3402
-0.0296-0.2091-0.0166-0.0135-0.20380.00620.0743
-0.0309-0.0210-0.26820.01480.0803
-0.1890
-0.0408-0.2291-0.1318
-0.1636-0.1768-0.1561-0.0137-0.0019-0.0128-0.0348
0.04360.4431
-0.1378
0.0163-0.0186-0.1624-0.1235-0.15210.0702
-0.18420.01430.54550.1110
-0.1384-0.1527
-0.1549-0.17020.05660.11890.18400.2931
-0.16420.0057
-0.0042-0.0064
-0.0267-0.04560.08360.1850
-0.0462-0.19160.60090.16060.1282
-0.07880.00290.20150.02880.0319
-0.1314-0.0009
0.00880.0775
-0.0225-0.1170-0.02150.00810.00440.0821
-0.10200.0100
-0.00580.03590.06580.07500.0427
0.05450.1432
-0.01610.02310.0651
-0.02820.06140.00960.05480.0262
-0.0615-0.01190.04120.04000.00580.0046
-0.04260.00570.00600.0227
-0.0179-0.0941-0.07500.0145
-0.03680.01660.00350.05200.0160
-0.1564
-0.01350.0819
-0.0100
-0.00390.0034
-0.0535-0.0340-0.01790.04460.0330
-0.0523-0.02450.00490.03090.01320.0138
-0.00730.0112
-0.0071-0.02940.03030.0879
-0.0231-0.0111-0.0023-0.00200.0009
-0.0022-0.0050
-0.0125-0.0253-0.03460.02860.0304
-0.05390.0146
-0.0371-0.0060-0.01480.0350
-0.0043
-0.03830.0078
-0.0141-0.00650.00260.0016
-0.0340-0.01770.0329
-0.0241
Object
596599606625631653661670673
675679688699706716719720729739742773776782785789793
808
824832847863908
913925929944984985
10111020
10361065109411261131113911431147
Bus Class
TKK
Sa
SK
K
S
S
S
K
C
Sq
CghS
SqL
X
KT
CghSI
Cb
X
S
SqL
S
A
L
Sa
SS
Sr
S
SrS
SS
XkA
S
S
S
DeMeo Class
T
L
L
SwS
K
K
SL
Sw
L
C
S
CghS
S
SqL
Xc
K
T
CghSw
CbX
S
SrL
S
S
A
DSw
SS
D
Sa
S
SwSr
SrS
Xk
Sw
SSw
D
Sw
Slope
0.28360.1751
0.1701
0.41980.1695
0.1222
0.09100.07480.1012
0.26980.1343
0.1442
0.11720.1744
0.08610.1976
0.08560.25920.1825
0.10830.30270.1045
0.33680.1886
0.21910.1374
0.16930.01200.1270
0.20650.87460.68580.35930.1022
0.2074
0.63000.43770.1807
0.29290.1983
0.12170.23030.21850.39410.14800.31140.75470.2828
PC1'
0.03740.16260.26440.40900.1668
-0.1679-0.07120.20280.4082
-0.06840.2956
-0.5997-0.0720-0.56330.45450.1490
-0.12980.2035
-0.2526
-0.24840.1142
-0.18680.0608
-0.5012-0.32080.2352
0.08620.42660.08970.1657
-0.7688-0.02160.2196
0.37380.11170.2156
-0.6705-0.0410-0.07750.2683
0.39060.06720.03180.24100.1474
-0.0401
0.06540.0410
PC2'
-0.2216-0.1783-0.2600-0.1305-0.0136-0.1498-0.16980.0193
-0.3053
-0.0132-0.2743-0.01200.03660.0475
-0.1267-0.02870.0391-0.2531-0.1227-0.1032-0.2490-0.1188-0.0072
-0.0460-0.0605-0.0073
0.0915-0.3690
-0.04280.03850.1836
-0.1444-0.0639-0.0883-0.0501-0.30830.1555
0.02110.05790.0304
-0.0041-0.0003-0.1897-0.0503-0.06730.0462
-0.2636-0.0232
PC3'
-0.1808-0.0262-0.0666
0.20460.0322
-0.0035
-0.0149-0.0003-0.0949
0.0450-0.0309-0.18690.0116
-0.0812-0.02240.0772
-0.03860.0273
-0.1680-0.0032-0.1274-0.14600.1467
-0.2471-0.1676-0.0096
-0.0014-0.0526-0.08270.0531
0.5417-0.00450.1825
-0.01330.1472
-0.12510.40030.11180.26340.0328
-0.01570.1669
-0.09900.23840.19830.2062
-0.09780.1653
PC4'
0.00210.03940.01450.0687
-0.0242-0.0703
-0.04150.0244
-0.0627
0.02180.0426
-0.07250.04670.2141
-0.00320.0078
0.0036-0.0633-0.0181-0.03900.01770.02640.0671
0.00590.0007
-0.0123
0.0518-0.10610.07060.0338
0.02950.23640.00320.0133
-0.0539-0.0059-0.0513
-0.0973-0.03010.0114
0.0087-0.00860.04970.0130
-0.03620.00160.1332
-0.0255
PC5'
-0.0279-0.00410.0114
-0.00980.0373
-0.0072
-0.03670.01410.00200.00480.0108
-0.01070.00810.07260.0232
-0.0304
0.03460.01900.0019
-0.01130.00320.0473
-0.0250-0.06610.00720.0300
0.0239-0.0267-0.02510.0618
-0.05900.0957
-0.01640.0088
-0.00120.0008
-0.0021-0.05360.02630.0402
0.0310-0.0101-0.02120.0301
-0.0050-0.0152-0.01070.0267
Object Bus Class DeMeo Class Slope PC1' PC2' PC3' PC4' PC5'
K1148
1198120412281300132913321350
1374
143314591471149415081542
1565
1620164016421658
165916601662
166716851751
18071839184818581862186418661903190419161929
19431980203520422045206320642074208520992107 S 0.2481 0.0580 -0.0705 0.0931 -0.0806 0.0095
K
Sw
Sw
Sr
CgSqw
LS
Sq
SVw
D
Sqw
B
DS
S
S
S
S
SS
Sr
SwS
S
SqwS
SS
QSqSw
K
VSw
V
Sw
Sw
Xe
Sr
V
Sq
Sqw
SqL
Ch
0.11890.39040.28760.20480.10540.30860.15290.22570.1482
0.15630.32460.77800.3476
-0.0249
0.72100.07350.1637
0.15950.09050.2413
0.16350.1779
0.1815
0.30460.09410.11430.33800.05190.14350.15040.11320.05280.34530.1249
0.14830.39020.21930.27320.40710.08760.08000.18140.07000.25690.45620.04540.1207
-0.20240.3122
-0.06140.2914
-0.2599-0.1273
0.08130.3398
-0.3566-0.01070.3316
-0.3535-0.1320
-0.45530.1271
-0.0435
-0.02880.0811
0.44150.5240
0.35040.22590.4950
0.2717-0.05660.3150
-0.22850.1526
0.24040.5822
-0.4924-0.27060.0700
-0.0241
0.2755-0.01241.0772
0.1740-0.01440.1472
0.1890
0.7055-0.3144-0.18390.1019
0.4460-0.5875
-0.1099-0.14380.01010.0708
-0.08180.0168
-0.3061-0.11710.1170
0.07121.3060
-0.06730.0756
-0.0766-0.19030.08900.0273
-0.0334-0.0859-0.1647-0.0797-0.0131
-0.0014-0.06190.0256-0.11470.1384
0.0026-0.0564-0.26250.1862
0.1082
-0.0057-0.1681
0.52570.05650.9150
-0.07460.0523-0.2595
0.07360.84180.06600.01950.0440
-0.35480.0166
0.0499-0.0191
0.10760.0311
-0.08370.1732
-0.13310.06410.1614
0.0725-0.2959-0.16560.2302
-0.1418-0.1536-0.19310.1421
0.07520.03000.0594
-0.04590.00620.08520.23160.1330
0.00200.07380.03230.07160.00290.22290.1794
0.0700-0.0055
-0.02710.0945
-0.00680.07760.1891
-0.0942-0.0102
-0.09650.1375
0.1767
0.1746
-0.0183
-0.1875
-0.06150.08270.04130.03190.0223
-0.0082-0.3096-0.12380.00020.0117
-0.22690.0479
-0.0403-0.06900.1144
-0.16280.05050.0067
-0.0789
0.07600.04920.0255
-0.00940.0180
-0.0230-0.09450.0275
-0.1107-0.10980.0101
-0.0743-0.0888-0.0052-0.0425
-0.08740.01450.03640.05240.0235
-0.0104-0.0016
0.0756-0.06530.00410.0269
-0.04880.0193
0.0004-0.0193-0.01860.03980.08340.0150
-0.08670.08030.00580.0271
-0.12110.00600.0252
-0.00880.01210.1260
-0.0231-0.03530.0501
-0.1031-0.03120.03370.06100.0174
-0.0037-0.0117-0.0119-0.01740.07260.0387
-0.01330.06190.0510
-0.0199
-0.0654-0.0035-0.02330.00610.03730.03000.0046
-0.06840.02260.0066
-0.0534-0.0129-0.0002
Object Bus Class DeMeo Class Slope PC1' PC2' PC3' PC4' PC5'
215722462335
2353
235423782386
23962401
24422448250125042521
256625792715
27322851287328752911
29122957
29652977
3028310231033122
31553198319932003248
32553317
3363339534023430
349135113628
3635370137343753 Q 0.2115 -0.5643 0.2866 0.0835 0.0046 -0.0157
S
L
CghS
Sa
S
L
A
Sq
S
V
V
A
A
V
SqS
S
VK
S
K
Xe
SV
S
SqB
D
ST
Sq
Sr
Sq
SqS
O
S
SLd
S
D
S
S
L
CghS
S
S
V
L
A
Sr
S
V
V
Sw
L
V
SqS
Sw
VK
Sv
S
K
Sqw
Xe
QwV
SqwK
B
D
S
D
Sr
S
S
S
S
Srw
O
SrwSL
0.18970.5312
0.2315
0.0935
0.07750.05830.1774
0.18490.17440.15710.31250.84370.11260.0789
0.0441
0.0889
0.31180.2002
0.1538
0.1345
0.1293
0.26790.1982
0.1449
0.1904
0.1668
0.01030.2604
0.1303
0.38290.0533
0.3354
0.2273-0.1626
0.48340.09950.5051
0.1822
0.1660
0.2381
0.0860
0.0196
0.32460.05720.27670.12250.0807
0.1990
0.2511
0.2507
0.3713
0.6610-0.54950.2766
0.2953
0.27660.88760.4253
-0.62650.0951
0.0866
0.7136
0.3419
0.32350.71630.6727
-0.17520.29650.22090.60010.01730.6981
0.2776
-0.1513-0.22800.2177
-0.3772
1.1248
-0.1666-0.6353-1.0224
0.34390.1948
-0.0303
-0.1342
0.1479
0.24310.2024
0.14750.0134
-0.79000.1795
0.26050.7779
-0.0655
-0.2110
-0.0416
-0.0718
-0.35590.0086
-0.0508
-0.1191
0.02000.9743
-0.30120.0916
0.0954-0.0090
0.84790.8659
-0.1613-0.3427
0.96080.1131
-0.0846-0.00711.1117
-0.15660.0147
-0.0690
-0.13960.0881
-0.2795
0.1962
0.8416
0.04900.04810.1534
-0.27830.0432
-0.1559
0.1633
0.0625-0.06620.0258
0.05070.11810.56960.1094
-0.0192
-0.4665
0.0722
-0.13410.0520
-0.0252
0.0723-0.1693
0.04220.1750
-0.0249
-0.2123
-0.0427
0.6472
-0.02930.0278
-0.1093
0.2123
0.22320.1762
-0.1887
0.1473
0.06670.0910
-0.2085-0.0374
-0.02330.0350
-0.0696
0.1889
-0.08290.2076
-0.3899
0.2175
0.1171-0.2026
-0.1477
0.0164-0.1150
0.0017
0.04210.1002
-0.0155
-0.04190.0887
0.3075
0.0258-0.01320.0210
-0.0001
0.1233
-0.0771
0.0137
-0.0512-0.00340.0113
-0.0343
-0.0367
-0.01000.06060.1271
0.0582
-0.0296
0.0515
0.0075
-0.0447-0.1555
-0.22130.0673
-0.04630.0362
-0.0381
-0.01210.0985
-0.0031
-0.1012-0.0515
-0.06940.0580
-0.1176
0.1583
0.0337
-0.12870.0552
0.01180.1130
0.02430.0495
-0.0218
0.0458-0.02310.00180.1764
0.0407-0.0652-0.0144
0.0245
0.0023
0.0458
0.0303
0.00240.02880.0136
0.0169-0.0399-0.0243
0.0336-0.08030.0279
-0.0044
-0.0638
-0.1497
-0.00860.02950.0734
-0.0298
0.01420.01210.0291
-0.02540.0146
0.0122
-0.0266-0.0098
0.0452-0.01150.1033
-0.08990.0088
-0.01230.01610.0365
-0.0036
0.04200.0724
-0.03680.0283
0.01990.0191
-0.10590.00550.0051
-0.0224
Object Bus Class DeMeo Class
3788 S S3844 L L3858 Sa Srw3873 S Sw3903 Sq S3908 V3910 S S
3920 Sa Sqw3949 Sq Sq4038 Vw4055 V4142 A L4179 Sk Sq4188 V V
4197 Sq Sq4352 S S
4407 Sa Sqw4417 S Sw4451 Sv4558 S Sr4570 Sa S4688 Q4713 A Sw4737 L L4995 S S
5013 SI Sw5111 R V
5143 O O
5230 S S
5261 Sr Sa5401 S S
5407 Sk S5587 Sq Sr5604 V
5641 A Sw5660 Q Q5685 S S
5817 S Sr5840 Ld L6047 S S
6239 Sqw
6386 S S
6411 B6455 S Srw6585 Sk S7341 Sq Q7763 L L8334 S S
Slope
0.20120.00260.26800.26650.0416
-0.01880.22820.34590.17610.25290.02880.10050.15620.16600.07680.1678
0.37860.25410.24200.1245
0.24400.21320.4111
0.06650.12580.33300.20660.06240.17240.48760.2466
0.06100.1093
-0.00590.41330.16840.1178
0.1528
0.06150.15830.28280.2317
-0.12750.32910.11470.07730.03690.1609
PC1'
0.2111
0.6056-0.01630.0244
-0.01450.59030.1284
-0.1114-0.3090
1.14410.33190.3651
-0.14300.4935
-0.09440.0187
-0.1573-0.01850.59060.1072
0.2939-0.85390.78320.71750.19290.01190.5694
-0.78420.0673
-0.9836
0.33630.1156
-0.03870.64290.2838
-0.70800.1648
0.29900.83210.1579
-0.32660.1362
-0.29720.40610.0539
-0.45100.54980.0801
PC2'
-0.0168-0.37880.14000.04830.04221.3071
-0.00080.07200.10600.29261.1709
-0.41770.09740.82540.08910.0446
-0.0039-0.0066-0.02210.0897
-0.11580.3441
-0.2321-0.39290.0414
-0.06800.59800.26190.04930.0731
-0.0713
0.01720.14040.8645
0.00660.2413
-0.03820.0282
-0.43940.02140.1338
-0.0010-0.11490.1819
0.05860.1878
-0.3018-0.0140
PC3'
0.0719-0.03300.32090.0577
-0.0252-0.30460.08660.21980.1596
-0.1791-0.26450.04780.0408
-0.1066-0.0621
0.1405
0.22660.1849
-0.0264-0.0154
0.16080.1742
0.0805-0.0188-0.03370.2286
-0.13070.12890.04940.24960.0723
-0.0276-0.0431-0.1587
0.19990.22130.0419
0.06970.0799
-0.07540.1199
0.1592-0.19560.20230.01180.1417
0.02340.0857
PC4'
0.0036-0.0933-0.19410.0638
-0.01710.0383
-0.0237-0.00880.02150.1398
-0.1467-0.38670.04160.1007
-0.0056
-0.0021-0.0549-0.03710.0122
-0.0412-0.0636-0.0777-0.0750-0.0332-0.01160.0571
-0.01100.0385
-0.0373-0.12720.05970.07130.0934
-0.0781-0.0199-0.1278-0.0482
-0.0362-0.1078-0.14870.08340.0515
-0.04240.22760.0238
-0.0197-0.02940.0747
PC5'
-0.0033-0.04270.1521
-0.00900.01910.08460.02390.02150.0086
-0.0909-0.0110-0.0015-0.0098-0.02080.0326
-0.0004
0.01090.0033
-0.00380.00290.00750.09870.03880.0214
-0.00020.0177
0.02670.01050.03380.08170.01950.0066
-0.00070.0369
-0.0020-0.0252-0.00080.0398
-0.0157-0.0642-0.0296-0.0416
0.0111-0.00380.0360
-0.0155-0.01880.0277
Object Bus Class DeMeo Class Slope PC1' PC2' PC3' PC4' PC5'
8444 S 0.1777 0.2300 -0.0559 0.1410 -0.0224 0.0044
17274 X 0.4237 -0.3824 -0.0526 -0.2156 -0.0445 -0.0349
18736 Sw 0.2492 0.0387 0.0908 0.0723 0.1699 -0.0230
19127 Qw 0.2832 -0.3785 0.2522 -0.0087 -0.1720 -0.1393
19356 S Sq 0.1229 -0.1385 0.1343 0.1090 -0.0537 0.0019
20786 Sq 0.0695 -0.3133 0.0392 -0.2523 -0.0629 -0.0224
20790 S 0.0486 0.2985 -0.0060 -0.0350 0.0465 -0.0096
22771 S 0.2264 0.6273 -0.1888 0.0128 0.0341 -0.0321
24475 Sw 0.2774 0.7547 -0.2018 0.1522 0.0178 -0.0313
35107 Sq 0.2409 -0.3046 0.1300 0.1665 0.0219 0.0273
36284 K 0.1107 -0.1968 -0.0413 0.0979 0.0223 -0.0324
53435 Srw 0.3577 0.3091 0.0856 -0.0179 0.0047 0.0077
54690 S 0.2020 0.5061 -0.2364 -0.0731 0.0468 0.0010
66146 Q 0.0858 -0.5855 0.1891 0.1026 -0.0367 -0.0099
86450 L 0.1180 0.3649 -0.2185 -0.0743 0.1205 -0.0517
86819 Sq 0.1680 -0.2950 0.1200 0.2179 -0.0268 -0.0452
89355 Sr 0.2111 0.2369 0.0799 0.0610 0.0568 0.0161
98943 Sw 0.3928 0.2022 -0.0203 0.0259 0.1419 -0.0436
137062 Sq Sr 0.1011 0.3418 0.2032 0.2097 0.2369 0.0584
1989VA Sq Sr 0.1541 -0.1192 0.2803 0.1263 0.1758 0.0530
1997AE12 Q 0.1561 -0.5143 0.2280 0.1113 -0.0834 -0.0288
2000GD2 Q 0.0732 -0.3689 0.1455 0.0965 0.0418 -0.0597
2000PG3 D 0.4300 0.3343 -0.2655 -0.2128 0.0289 -0.0332
2000XL44 S 0.2240 -0.0382 0.0331 0.1544 0.0537 -0.0624
2001MQ3 K 0.1459 0.0640 -0.1642 0.0521 -0.2369 0.0169
2001TX16 X 0.2436 -0.0120 -0.1664 -0.2179 0.0668 -0.0506
2001XN254 S 0.1939 0.3016 -0.0733 0.0756 -0.0163 -0.0943
2002AA S 0.0267 0.1247 0.0117 -0.0773 -0.0387 -0.0345
2002AV S 0.1646 -0.0534 -0.0110 -0.0969 -0.0110 -0.0641